# Computations
import pandas as pd
import numpy as np

# Visualisation libraries

## Text
from colorama import Fore, Back, Style
from IPython.display import Image, display, Markdown, Latex, clear_output

## plotly
from plotly.offline import init_notebook_mode, iplot 
import plotly.graph_objs as go
import plotly.offline as py
from plotly.subplots import make_subplots
import plotly.express as px

## seaborn
import seaborn as sns

## matplotlib
import matplotlib.pyplot as plt
from matplotlib.patches import Ellipse, Polygon
from matplotlib.font_manager import FontProperties
import matplotlib.colors as mcolors
plt.style.use('seaborn-whitegrid')
plt.rcParams['axes.labelsize'] = 14
plt.rcParams['xtick.labelsize'] = 12
plt.rcParams['ytick.labelsize'] = 12
plt.rcParams['text.color'] = 'k'
%matplotlib inline

import warnings
warnings.filterwarnings("ignore")

Home / Business Analytics and Modeling

Bank Marketing Dataset

• Preprocessing
• Exploratory Data Analysis
• Modeling: PyTorch Multi-layer Perceptron (MLP) for Binary classification
• Modeling: Keras Multi-layer Perceptron (MLP) for Binary classification

Bank Marketing Dataset

In this article, we work on a dataset available from the UCI Machine Learning Repository. The data is related to direct marketing campaigns (phone calls) of a Portuguese banking institution. The classification goal is to predict if the client will subscribe to a term deposit (variable y).

This dataset is based on "Bank Marketing" UCI dataset (please check the description at archive.ics.uci.edu/ml/datasets/Bank+Marketing). The data is enriched by the addition of five new social and economic features/attributes (national wide indicators from a ~10M population country), published by the Banco de Portugal and publicly available at: bportugal.pt/estatisticasweb. This dataset is almost identical to the one used in [Moro et al., 2014] (it does not include all attributes due to privacy concerns).

Dataset Information:

The data is related to the direct marketing campaigns of a Portuguese banking institution. The marketing campaigns were based on phone calls. Often, more than one contact to the same client was required, in order to access if the product (bank term deposit) would be ('yes') or not ('no') subscribed.

There are four datasets:

1. bank-additional-full.csv with all examples (41188) and 20 inputs, ordered by date (from May 2008 to November 2010), very close to the data analyzed in [Moro et al., 2014]
2. bank-additional.csv with 10% of the examples (4119), randomly selected from 1), and 20 inputs.
3. bank-full.csv with all examples and 17 inputs, ordered by date (older version of this dataset with fewer inputs).
4. bank.csv with 10% of the examples and 17 inputs, randomly selected from 3 (older version of this dataset with fewer inputs).

The classification goal is to predict if the client will subscribe (yes/no) a term deposit (variable y).

Loading the Dataset

The zip file includes two datasets:

1. bank-additional-full.csv with all examples, ordered by date (from May 2008 to November 2010).
2. bank-additional.csv with 10% of the examples (4119), randomly selected from bank-additional-full.csv. The smallest dataset is provided to test more computationally demanding machine learning algorithms (e.g., SVM).

The binary classification goal is to predict if the client will subscribe a bank term deposit (variable y).

Data = pd.read_csv('Data/bank-additional-full_EDA.csv')
display(Data.head(10).style.hide_index().set_precision(2))
Target = 'Term Deposit Subscription'
Labels = ['No', 'Yes']

Age	Job	Marital	Education	Default	Housing	Loan	Contact	Month	Day Of Week	Duration	Campaign	Pdays	Previous	Poutcome	Employment Variation Rate	Consumer Price Index	Consumer Confidence Index	Euribor three Month Rate	Number of Employees	Term Deposit Subscription
56	Housemaid	Married	Basic 4Y	No	No	No	Telephone	May	Monday	261	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
57	Services	Married	High School	Unknown	No	No	Telephone	May	Monday	149	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
37	Services	Married	High School	No	Yes	No	Telephone	May	Monday	226	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
40	Admin.	Married	Basic 6Y	No	No	No	Telephone	May	Monday	151	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
56	Services	Married	High School	No	No	Yes	Telephone	May	Monday	307	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
45	Services	Married	Basic 9Y	Unknown	No	No	Telephone	May	Monday	198	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
59	Admin.	Married	Professional Course	No	No	No	Telephone	May	Monday	139	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
41	Blue-Collar	Married	Unknown	Unknown	No	No	Telephone	May	Monday	217	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
24	Technician	Single	Professional Course	No	Yes	No	Telephone	May	Monday	380	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No
25	Services	Single	High School	No	Yes	No	Telephone	May	Monday	50	1	999	0	Nonexistent	1.10	93.99	-36.40	4.86	5191.00	No

Number of Instances	Number of Attributes
41188	21

Bank Client Data

Feature	Description
Age	numeric
Job	Type of Job (categorical: "admin.","blue-collar","entrepreneur","housemaid","management","retired","self-employed","services","student","technician","unemployed","unknown")
Marital	marital status (categorical: "divorced","married","single","unknown"; note: "divorced" means divorced or widowed)
Education	(categorical: "basic.4y","basic.6y","basic.9y","high.school","illiterate","professional.course","university.degree","unknown")
Default	has credit in default? (categorical: "no","yes","unknown")
Housing	has housing loan? (categorical: "no","yes","unknown")
Loan	has personal loan? (categorical: "no","yes","unknown")

Related with the Last Contact of the Current Campaign

Feature	Description
Contact	contact communication type (categorical: "cellular","telephone")
Month	last contact month of year (categorical: "jan", "feb", "mar", ..., "nov", "dec")
Day of week	last contact day of the week (categorical: "mon","tue","wed","thu","fri")
Duration	last contact duration, in seconds (numeric). Important note: this attribute highly affects the output target (e.g., if duration=0 then y="no"). Yet, the duration is not known before a call is performed. Also, after the end of the call y is obviously known. Thus, this input should only be included for benchmark purposes and should be discarded if the intention is to have a realistic predictive model.

Other Attributes

Feature	Description
Campaign	number of contacts performed during this campaign and for this client (numeric, includes last contact)
Pdays	number of days that passed by after the client was last contacted from a previous campaign (numeric; 999 means client was not previously contacted)
Previous	number of contacts performed before this campaign and for this client (numeric)
Poutcome	outcome of the previous marketing campaign (categorical: "failure","nonexistent","success")

Social and Economic Context Attributes

Feature	Description
Employment Variation Rate	employment variation rate - quarterly indicator (numeric)
Consumer Price Index	consumer price index - monthly indicator (numeric)
Consumer Confidence Index	consumer confidence index - monthly indicator (numeric)
Euribor three Month Rate	euribor* 3 month rate - daily indicator (numeric)
Number of Employees	number of employees - quarterly indicator (numeric)

* the basic rate of interest used in lending between banks on the European Union interbank market and also used as a reference for setting the interest rate on other loans.

Output variable (Desired Target):

Feature	Description
Term Deposit Subscription	has the client Term Deposit Subscription? (binary: "yes","no")

Dataset_Subcategories = {}
Dataset_Subcategories['Bank Client Data'] = Data.iloc[:,:7].columns.tolist()
Dataset_Subcategories['Related with the Last Contact of the Current Campaign'] = Data.iloc[:,7:11].columns.tolist()
Dataset_Subcategories['Other Attributes'] = Data.iloc[:,11:15].columns.tolist()
Dataset_Subcategories['Social and Economic Context Attributes'] = Data.iloc[:,15:-1].columns.tolist()
Dataset_Subcategories['Output variable (Desired Target)'] = Data.iloc[:,-1:].columns.tolist()

Exploratory Data Analysis

def FeatAgg(Feat, ColorFeat, Target = Target, Inp = Data):
    Out = Inp[[Feat, ColorFeat,Target]]
    Out = Out.groupby([Feat, ColorFeat,Target])[Target].agg({'count'}).rename(columns = {'count':'Count'})
    Out['Percentage'] = np.round(100* Out.values /Out.sum().values, 2)
    Out.reset_index(drop = False, inplace = True)
    Out = Out.sort_values(by=[Feat])
    Out[Feat] = Out[Feat].astype(str)
    return Out

def DistPlot(Feat, Target = Target, nbins = 20,
             Colors = ['LightSalmon', 'LightBlue'], LC = 'Black',
             yLim = [0, 80], H = 450, titleY = 0.92, Inp = Data):
    fig = px.histogram(Inp, x = Feat, nbins=nbins, color= Target, marginal= 'box',
                       color_discrete_sequence= Colors, hover_data=Data.columns)
    fig.update_xaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                     zeroline=False, zerolinewidth=1, zerolinecolor='Black',
                     showgrid=False, gridwidth=1, gridcolor='Lightgray')
    fig.update_yaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                     zeroline=True, zerolinewidth=1, zerolinecolor='Black',
                     showgrid=True, gridwidth=1, gridcolor='Lightgray')

    Name = '%s Distribution by %s' % (Target, Feat)

    fig.update_layout(legend_orientation='v', plot_bgcolor= 'white', height= H, width= 980,
                     title={'text': '<b>' + Name + '<b>', 'x':0.5, 'y': titleY, 'xanchor': 'center', 'yanchor': 'top'},
                     yaxis_title='Frequency')
    fig.update_traces(marker_line_color= LC, marker_line_width=0.5, opacity=1)
    fig['layout']['yaxis'].update(range=yLim)
    fig.show()

# For plotting
def FeatCut(Feat, ColorFeat, Bins, Target = Target, Inp = Data):
    Out = Inp[[Feat, ColorFeat, Target]]
    Out[Feat] = pd.cut(Out[Feat], bins = pd.IntervalIndex.from_tuples([(x, y) for x, y in zip(Bins[:-1],Bins[1:])]))

    Out = Out.groupby([Feat, ColorFeat,Target])[Target].agg({'count'}).rename(columns = {'count':'Count'})
    Out['Percentage'] = np.round(100* Out.values /Out.sum().values, 2)
    Out.reset_index(drop = False, inplace = True)
    Out = Out.sort_values(by=[Feat])
    Out[Feat] = Out[Feat].astype(str)
    return Out

def PlotX(df, Feat, ColorFeat, Target = Target,
           Colors = list(mcolors.TABLEAU_COLORS.values()), LC = 'Black',
           yLim = [0, 35], H = 500, titleY = 0.90):   
    # Figure
    fig = make_subplots(rows=1, cols=2, horizontal_spacing = 0.02, shared_yaxes=True, y_title = 'Percent',
                        subplot_titles=('%s: <b>No<b>' % Target, '%s: <b>Yes<b>' % Target))
    # Left
    if Colors == None:
        fig1 = px.bar(df.loc[df[Target] == 'No'], x=  Feat, y= 'Percentage', orientation='v',
                      color = ColorFeat, text = 'Percentage', hover_data= df.columns)
    else:
        fig1 = px.bar(df.loc[df[Target] == 'No'], x=  Feat, y= 'Percentage', orientation='v',
                      color = ColorFeat, text = 'Percentage', hover_data= df.columns,
                      color_discrete_sequence = Colors)

    for i in range(len(fig1['data'])):
        fig.add_trace(fig1['data'][i], row=1, col=1)
    fig.update_traces(marker_line_color= LC, marker_line_width=1, opacity=1, row=1, col=1)

    # Right
    if Colors == None:
        fig2 = px.bar(df.loc[df[Target] == 'Yes'], x= Feat, y= 'Percentage', orientation='v',
                      color = ColorFeat, text = 'Percentage', hover_data= df.columns)
    else:
        fig2 = px.bar(df.loc[df[Target] == 'Yes'], x= Feat, y= 'Percentage', orientation='v',
                      color = ColorFeat, text = 'Percentage', hover_data= df.columns,
                      color_discrete_sequence = Colors)

    for i in range(len(fig2['data'])):
        fig.add_trace(fig2['data'][i], row=1, col=2)
    fig.update_traces(marker_line_color= LC, marker_line_width=1, opacity=1, showlegend = False, row=1, col=2)

    # Update
    fig.update_xaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                     zeroline=False, zerolinewidth=1, zerolinecolor='Black',
                     showgrid=False, gridwidth=1, gridcolor='Lightgray')
    fig.update_yaxes(showline=True, linewidth=1, linecolor='Lightgray', mirror=True,
                     zeroline=False, zerolinewidth=1, zerolinecolor='Black',
                     showgrid=True, gridwidth=1, gridcolor='Lightgray', range= yLim)
    fig.update_layout(legend_orientation='v', legend_title_text=ColorFeat, plot_bgcolor= 'white', height= H, width= 980)
    fig.update_layout(legend=dict(font=dict(color="Black"), bordercolor="Lightgray", borderwidth=1))
    Name = '%s Distribution by %s and %s' % (Feat, ColorFeat, Target)
    fig.update_layout(title={'text': '<b>' + Name + '<b>', 'x':0.5, 'y': titleY, 'xanchor': 'center', 'yanchor': 'top'})
    fig.show() 

Age

Feat = 'Age'
DistPlot(Feat, yLim = [0, int(1e4)], Colors = ['LightCoral', 'LawnGreen'])

Bins = [14, 24, 40, 59, 80, 100]
ColorFeat = 'Marital'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 30])

ColorFeat = 'Default'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 50], Colors = None)

ColorFeat = 'Housing'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 30], Colors = None)

ColorFeat = 'Loan'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 50], Colors = None)

ColorFeat = 'Contact'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 35], Colors = None)

ColorFeat = 'Poutcome'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 50], Colors = None)
del ColorFeat, Table, Bins

Job

Feat = 'Job'
DistPlot(Feat, yLim = [0, int(1e4)], Colors = ['LightCoral', 'LawnGreen'])

ColorFeat = 'Marital'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)

ColorFeat = 'Default'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)

ColorFeat = 'Housing'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 14], Colors = None)

ColorFeat = 'Loan'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)

ColorFeat = 'Contact'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 15], Colors = None)

ColorFeat = 'Poutcome'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)
del ColorFeat, Table

Education

Feat = 'Education'
DistPlot(Feat, yLim = [0, int(15e3)], Colors = ['LightCoral', 'LawnGreen'])

ColorFeat = 'Marital'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 14], Colors = None)

ColorFeat = 'Default'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 25], Colors = None)

ColorFeat = 'Housing'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 14], Colors = None)

ColorFeat = 'Loan'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 25], Colors = None)

ColorFeat = 'Contact'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)

ColorFeat = 'Poutcome'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 25], Colors = None)
del ColorFeat, Table

Housing

ColorFeat = 'Marital'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 14], Colors = None)

ColorFeat = 'Default'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 25], Colors = None)

ColorFeat = 'Loan'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 25], Colors = None)

ColorFeat = 'Contact'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)

ColorFeat = 'Poutcome'
Table = FeatAgg(Feat, ColorFeat)
PlotX(Table, Feat, ColorFeat, yLim = [0, 25], Colors = None)
del ColorFeat, Table

Campaign

Feat = 'Campaign'
DistPlot(Feat, yLim = [0, int(4e4)], Colors = ['LightCoral', 'LawnGreen'], nbins = 15)

Bins = [0, 5, 10, 20, 50, 100]
ColorFeat = 'Marital'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 50])

ColorFeat = 'Default'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 70], Colors = None)

ColorFeat = 'Housing'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 50], Colors = None)

ColorFeat = 'Loan'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 80], Colors = None)

ColorFeat = 'Contact'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 60], Colors = None)

ColorFeat = 'Poutcome'
Table = FeatCut(Feat, ColorFeat, Bins)
PlotX(Table, Feat, ColorFeat, yLim = [0, 80], Colors = None)
del ColorFeat, Table, Bins

Duration

Feat = 'Duration'
DistPlot(Feat, yLim = [0, int(4e4)], Colors = ['LightCoral', 'LawnGreen'])

Bins = [-1, 100, 200, 400, 800, 1000, 5000]
ColorFeat = 'Marital'
Table = FeatCut(Feat, ColorFeat, Bins).replace({'(-1, 100]':'[0, 100]'})
PlotX(Table, Feat, ColorFeat, yLim = [0, 20])

ColorFeat = 'Default'
Table = FeatCut(Feat, ColorFeat, Bins).replace({'(-1, 100]':'[0, 100]'})
PlotX(Table, Feat, ColorFeat, yLim = [0, 30], Colors = None)

ColorFeat = 'Housing'
Table = FeatCut(Feat, ColorFeat, Bins).replace({'(-1, 100]':'[0, 100]'})
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)

ColorFeat = 'Loan'
Table = FeatCut(Feat, ColorFeat, Bins).replace({'(-1, 100]':'[0, 100]'})
PlotX(Table, Feat, ColorFeat, yLim = [0, 30], Colors = None)

ColorFeat = 'Contact'
Table = FeatCut(Feat, ColorFeat, Bins).replace({'(-1, 100]':'[0, 100]'})
PlotX(Table, Feat, ColorFeat, yLim = [0, 20], Colors = None)

ColorFeat = 'Poutcome'
Table = FeatCut(Feat, ColorFeat, Bins).replace({'(-1, 100]':'[0, 100]'})
PlotX(Table, Feat, ColorFeat, yLim = [0, 30], Colors = None)
del ColorFeat, Table, Bins

---

Refrences

1. S. Moro, P. Cortez and P. Rita. A Data-Driven Approach to Predict the Success of Bank Telemarketing. Decision Support Systems, Elsevier, 62:22-31, June 2014

2. S. Moro, R. Laureano and P. Cortez. Using Data Mining for Bank Direct Marketing: An Application of the CRISP-DM Methodology. In P. Novais et al. (Eds.), Proceedings of the European Simulation and Modelling Conference - ESM'2011, pp. 117-121, Guimaraes, Portugal, October, 2011. EUROSIS. [bank.zip]

---